Pay for Performance? It Depends on the Measuring Stick
Year after year, as
executive pay continues its
inexorable climb, it’s amusing to watch corporate directors try to
justify the piles of shareholder money they throw at the hired help.
Check out any proxy filing for these arguments, which usually center
on how closely and carefully the executives’ incentive compensation is
tied to the performance of company operations.
But pay for performance
is only as good as the metrics used to determine it. And as a
recent study shows, some metrics
— including the most popular — are downright ineffective at motivating
executives to create shareholder value.
The study was done by
James F. Reda, a veteran compensation consultant, and his associate
David M. Schmidt, both of whom are in the human resources and
compensation consulting practice at
Arthur J. Gallagher & Company.
They analyzed pay metrics used by 195 large companies over the five
years that ended in December 2012. By comparing those measurements
with moves in these companies’ stock prices, the study identified the
common pay metrics that corresponded with above- or below-average
Credit Minh Uong/The
New York Times
Their analysis will come
in handy for investors examining the executive-pay tallies for 2013.
As usual, the numbers are staggering: The median compensation for
C.E.O.s at the 100 largest companies that have filed so far was $13.9
million, according to the Equilar 100 C.E.O. Pay Study, conducted by
Equilar, an executive compensation data firm. That’s up 9 percent from
But investigating the
basis of these amounts takes some digging. Consider Oracle, whose
chief executive, Lawrence J. Ellison, received $78.4 million, placing
him atop our 2013 pay list. Only by reading the company’s proxy do you
learn that Oracle determined its incentive compensation — meaning most
everything but salary — based on growth in what it calls “non-GAAP
In Oracle’s lexicon, that
means the company’s earnings before income taxes and minus the costs
of stock-based compensation, acquisitions, restructurings and other
items. In other words, the Oracle number is not based on generally
accepted accounting principles, or GAAP, and that makes its numbers
Oracle’s approach is just
one of many benchmarks companies can choose. Some boards award
incentive pay based on a company’s total shareholder return or
earnings-per-share growth; others use return on invested capital or
return on equity. Most companies use more than one measure. And all
argue that their methods justify the incentive pay they award.
But which measurements
work and which don’t?
According to Mr. Reda and
Mr. Schmidt, stocks of companies choosing the most popular gauge —
total shareholder return — as a performance metric for at least one
year out of the five in the study underperformed stocks of companies
using other benchmarks. By contrast, stocks of companies that used
earnings-per-share measures based on generally accepted accounting
The analysis also
determined that companies making frequent changes to their pay metrics
vastly underperformed those that stuck with their benchmarks.
It is dismaying that
companies using total shareholder return as a performance metric
tended to underperform, given its rising popularity in pay practices.
Among the 195 companies in the study, just over half — 53 percent —
used total shareholder return as a metric. Those companies’ shares had
an average loss of 0.18 percent, annualized, over the five-year
period. That compares with an average gain of 1.15 percent among all
195 stocks, regardless of the benchmark they used.
Even more striking,
stocks of companies that did not use total shareholder return as a
measure gained an annualized average of 2.67 percent.
“What this says is
companies should look for alternatives to the total shareholder return
measure,” Mr. Reda said. “They should look for growth drivers that
management can control and not do the knee-jerk thing.” That knee-jerk
thing, he said, is total shareholder return, and he said that using it
routinely was a mistake.
Interestingly, the study
found that stocks of companies using a much less popular metric —
earnings-per-share growth — were more likely to outperform. Only 37
percent of companies used that measure at least once from 2008 to
2012, the study found. Still, these shares generated a gain of 1.37
percent, annualized, during the five years, above the 1.15 percent
average gain across all the companies.
Neither Mr. Reda nor Mr.
Schmidt could say with certainty why the benchmark of total
shareholder return seemed so closely linked to lackluster corporate
performance. But Mr. Reda was willing to speculate on why
earnings-per-share measures were less popular in the boardroom. They
are harder to manipulate than other measures that burnish results by
removing costs from the equation, he said. (Of course, earnings can be
enhanced by stock buybacks and other management machinations, but most
of those tricks can be spotted fairly easily.)
“Earnings per share seems
to be the best measure, and lots of investors and shareholders think
it is important,” Mr. Reda said. “But companies hate it. They don’t
want to be held accountable for the costs of discontinuing product
lines or closing factories.”
Another conclusion from
the study is this: Shares of companies that chose a metric and stuck
with it generally did better than those of companies that changed
their measures. For example, the 24 companies that changed their
performance measures three times over the five years generated an
average loss of 2.65 percent, annualized. The 56 companies that
maintained the same metrics over the period — it didn’t matter which
metric they used — showed gains of 5.09 percent.
“Stability of design has
value,” Mr. Reda said.
Perhaps the study’s
clearest message is that one size does not fit all when measuring pay
for performance. For instance, not all companies’ stocks
underperformed when their managers were judged on total shareholder
Health care equipment and
services companies are an example. The returns of those that used
total shareholder return exceeded the average returns of their peers,
the study found. Of the 11 companies studied in that industry, the
stocks of the three using total shareholder return generated average
annualized gains of 4.4 percent in the period. That compares with 1.9
percent for shares of companies that didn’t use the metric.
Similarly, the use of
earnings per share didn’t always translate to stock outperformance,
the study found. This was the case in the energy,
insurance, media and retailing
As a result, Mr. Reda
said, boards must design their pay packages with goals that are
specific to the company’s strengths and weaknesses, but that will also
promote the kind of growth that shareholders want.
“Large companies can be
unwieldy,” Mr. Reda said. “If management is not really focused on what
they need to do, they are unlikely to succeed in getting their houses
in order. If you can focus management on things they can control,
investors might be better off.”
version of this article appears in print on April 13, 2014, on page
BU1 of the New York edition with the headline: Pay for Performance? It
Depends on the Measuring Stick
© 2014 The
New York Times Company